Communication of Hurricane Storm Surge Threat: A Mixed-Method Analysis

Communication of Hurricane Storm Surge Threat: A Mixed-Method Analysis

Kathleen Sherman-Morris, Karla B. Antonelli,

Carrick C. Williams, Amanda Lea

• Many graphic styles used to communicate weather information.

• Little research has been done on understanding • Test effectiveness of storm surge graphics as a

function of color palette and legend type.

Purpose

Two Methods• Eye tracking experiment• MSU Eye tracking lab• Participants included Student and

General Public • Online survey of Mississippi and

Alabama Gulf Coast residents• SurveyMonkey• Email-mailout from InfoUSA,

supplemented by publicity from local TV meteorologist

• Targeted Harrison and Jackson Counties in MS, Mobile and Baldwin Counties in Alabama.

What does eye-tracking add?

• A real-time, online measure of behavior

• Eye movements can be seen as a stand-in for attention

• Measures of where and how long looking at stimuli can measure interest, effort to process

Apparatus

• ISCAN ETL-400• Uses a camera, infrared light shone into

subject’s right eye• Camera records reflection from cornea (CR)

and the lack of reflection from pupil• Calibrates these angles with computer display

to determine where subject is looking• Chin rest used for head stabilization

Participants

• 40 participants • Recruited from MSU meteorology program,

MSU general student body, and Starkville community

• 21 male, 19 female• Aged 19 to 55 years

Eye Tracking Procedure

• Within subjects design with 5 levels– 3 color palettes– 2 legends (feet and text)

• Order and location were varied for familiarity effects

• 8 questions for each image (4 accuracy, 2 behavioral, 1 risk perception rating and 1 helpfulness rating)

QuestionsQuestion 1 Which color do you believe is associated with the highest and

worst storm surge?Question 2 In which location is the storm surge forecast to be higher?Question 3 What is the storm surge at point A forecast to be?Question 4 Do you think a property located right at point A would

experience storm surge flooding?Question 5 If you lived in a single level house or on the ground level of an

apartment building at point A, would you take any precautions to prevent damage to your home or belongings?

Question 6 If you lived in a single level house or on the ground level of an apartment building at point B, would you take any precautions to prevent damage to your home or belongings?

Question 7 On a scale from 1 to 8, where 1 is not bad at all and 8 is very bad, how would you rate this hurricane based on its storm surge potential?

Question 8 On a scale from 1 to 8, where 1 is not helpful at all and 8 is very helpful, how helpful do you think this image is in your ability to judge the storm surge risk associated with this hurricane?

Image Conditions--Green/Red

** No legend on 1st image

Green/Red Values (Feet)

Yellow/Purple Text

Yellow/Purple Values (Feet)

Blue Values (Feet)

Blue Values G/R Text G/R Values Y/P Text Y/P Values30

31

32

33

34

35

Average accuracy, all hurricanesO

ut o

f 40

subj

ects

Which does the BEST job of informing the public about their storm surge risk?

Legend Type Color PaletteValues Text Blue G/R Y/P

Experts 7 2 2 5 2Community 9 11 2 14 3Undergraduates 5 6 1 7 3Total 21 19 5 26 8

Eye Tracking Results

• Blue had longest response time• Text references had somewhat shorter

response times than feet• Green/red produced most accurate* results,

but accuracy was high for all palettes• People thought the green/red palette was the

best, but did not prefer one legend type over the other

* Not significant in most cases

Online Survey, briefly

• 129 responses from Alabama counties• Average age 55 (range 19-84)• 40% women, 43% men, 7% no response• 76% White• 78% homeowners

Scenarios

• Same color palettes/legends• One stronger storm, one weaker storm• 8 risk perception questions, 1 accuracy

question, 1 intended behavior, 1 perceived helpfulness

Weaker storm results

Accuracy by palette and legend

Text Values (Feet)Green/Red 88.4% 85.0%Yellow/Purple 83.6% 66.7%Blue 73.5%

Online Survey Results

• No significant differences in risk perception among palettes and legends

• Legend values in feet led to marginally less accuracy than text

• Green/Red more accurate than Blue combined with Yellow/Purple– Could this be a legend effect?– More responses are necessary for further

evaluation

Acknowledgements

• This project was funded by NOAA, through the Northern Gulf Institute. The authors also thank Ethan Gibney, for creating the storm surge images.